Temperament and sexual behaviour in the Furrowed Wood Turtle Rhinoclemmys areolata.

The variation in temperament among animals has consequences for evolution and ecology. One of the primary effects of consistent behavioral differences is on reproduction. In chelonians some authors have focused on the study of temperament using different methods. In our research our first aim was i) establish a methodology to determine the degree of boldness among individuals Rhinoclemmys areolata. Our second aim was to ii) determine the role boldness plays during reproduction, with emphasis on courtship and copulation, considering a) the interactions between males and females, and b) competition between males. We used 16 sexually mature individuals of each sex. Males were observed in four different situations and 17 behavioral traits were recorded. We selected 12 traits that allowed us distinguish between the bolder and the shier individuals and found that five behavioral traits were specific for bolder individuals and five others for shier individuals. In a second step, we observed a male in presence of a female and recorded courtship behaviors and breeding attempts. Bolder individuals did not display courtship behaviors and just attempted to copulate. Shier individuals displayed courtship behaviors and copulation attempts were rarely observed. Finally, in the simulations that compared two males in the presence of a female we noticed that bolder individuals displayed courtship behaviors while the shier ones simply ignored the female. Our results first allowed us to determine which methodology is the best to determine temperament in turtles. Secondly, temperament seems to be an important factor in modulating interaction between males and females. Bolder individuals have an advantage during competition and display courtship behaviours only if other males are present. Shier males displayed courtship behaviors and only try to copulate when no competitors were present. These two different temperament-dependant strategies are discussed in terms of ecology, evolution and management.


Introduction
Inter-individual behavioral differences in an animal species are consistent or largely maintained over time and in diverse situations [1] and sometimes between one population and the other [2]. These differences are termed "temperament", "behavioral syndrome" or "reactivity" Recently, boldness has been studied in aquatic turtles. The recognition between males and the reduction of competitive interactions are related to the boldness of males [40]. They showed that the response to chemical stimuli of known or unknown males of the Mediterranian Turtle Mauremys leprosa, depends on the level of boldness, in particular the shiest males are those who avoid encountering other males, of the same or different species contributing to the stabilization of social systems and reducing the frequency and intensity of aggressive encounters between males. Different methods are used to evaluate temperament in turtles. Reactivity level is measured using the straightening response, which is the time it takes for an individual to become upright after being turned upside down on its carapace [41]. A similar method has been used to study anti-predator responses in the European pond turtle Emys orbicularis [42]. Another method consists in placing individuals in a new environment: here, the latency to move from an initial position in an arena was used to evaluate exploration in the red-eared slider turtle Trachemys scripta [43] and in the Eastern hermann Eurotestudo boettgeri [44]. The presentation of a threatening stimuli was used to measure reactivity and the effect of new objects on investigative behaviors while researching exploration in the Agassiz´s desert tortoise (Gopherus agassizii) [4]. Boldness in the Eastern box turtle (Terrapene carolina) was evaluated by recording the time taken to emerge from the shell and to move after a short period of confinement [45]. Evidently, there are diverse methods applicable to turtles; however, none of the aforementioned studies tried to apply more than one method to a single species. Therefore, a unique methodology for the study of boldness in turtles is still lacking.
Unfortunately, there is very little research that specifically addresses how temperament may be related to the reproductive behavior of turtles, particularly its influence on male competition behavior, courtship and coupling. Thus, the aims of this study were to (i) determine the temperament of Rhinoclemmys areolata and provide a method that not only ascertains temperament for this species but also other species of terrestrial turtle (ii) understand if its temperament could be a factor in modulating the interaction between male and female, and (iii) between males during competition for a female.

Ethical statement
The investigation was carried out in compliance with the institutional ethical standards and norms in force. All the animals come from private loans from the city of Chetumal (Quintana Roo-Mexico). Verbal consent was requested from the owners, and it was explained how the turtles would be used and for what objective. The individuals on loan were destined exclusively for specific research purposes, respecting all known needs for animal welfare, and were returned to their owners at the end of the data collection.
We were in possession of the permit N˚SGPA / DGVS / 002491/18 issued by Secretaria de Medio Ambiente y Recursos Naturales (SEMARNAT) for the collection of this species.
Experimental protocol was approved by Ethical Committee from the "El Colegio de la Frontera Sur", Mexico.

Object of study
For the experiment, we used 16 males and 16 females of the furrowed wood turtle Rhinoclemmys areolata, all sexually mature. This species occurs in Yucatán Peninsula, Cozumel Island, Northern Guatemala, Belize, Northwestern Honduras, where it inhabits savannah, thorny scrub, broad-leaved forests, swampy and gouache areas [46]. It is used as a pet, for food and traditional medicine [47]. It is listed as Near-threatened by the IUCN Red List [48]. It is also known as "Chak pool" or "Mojina" [47]. It is characterized by 1 to 2 red, orange or yellow stripes on the top of the head, and the plastron is yellow with a dark spot of variable size ( Fig  1). It presents a low level of sexual dimorphism, males having a longer and thinner tail than females, and also having a slightly more concave plastron; however, this is characteristic is not always detectable (Fig 2). It is a terrestrial species, whose only connection with the aquatic environment is during courtship and copulation, although this information solely derives from two anecdotic observations events where the individuals were kept in pools without the presence of a large terrestrial environment [49,50]. Courtship and copulation could be induced by spraying the individuals with water [46].

Turtles' management and setup
Individuals of Rhinoclemmys areolata were kept within a 16 x 5 m enclosure, located in the facilities of El Colegio de la Frontera del Sur (ECOSUR) Unidad Chetumal (18˚32'37.4''N, 881 5 ' 48.0''W). The area was fenced and divided into two parts of approximately the same size (5 x 8 m) containing male and female individuals, respectively. This division prevented any reproductive activity in the absence of the observer. It also ensured that the female did not store semen and therefore be less reactive to males. Females from some species of turtle are able to retain vital sperm for up to 4 years; however, it is not yet clear in which species, therefore we cannot dismiss that it could also occur in R. areolata [51].
The area inside the fence resembled as close as possible the natural habitat of R. areolata, by maintaining the presence of trees, foliage, natural soil and adding water sites (Fig 3).
Turtles were fed with fruit, vegetables and meat on alternate days. Inside the maintenance area, they were free to move, to find refuge and bathe in small pools. The water in the maintenance area was changed daily. They were subject to natural temperatures and rainfall for as long as they were inside the area in order to sustain optimum environmental conditions for their welfare.
All simulations were performed in a mobile enclosed area (2 x 2 m) (Fig 4). Each simulation was recorded through two cameras (YiLiteCam and Motorola Camera). The first video camera was positioned above the observation area while the second was moved manually by the observer in the event that details of specific events were required. There were no food or small pools in the experimental area that could have had an effect on turtle behavior [41].
Each simulation took place in the open air, on sunny days and during the hours of maximum turtle activity (8.30am-11.30am and 05.30pm-06.30pm) when temperatures were not very high.
Animals were separated after each observation session to avoid any additional intersexual interactions in absence of the observer. The observation area was moved to an adjacent area after each simulation to reduce the time elapsed between experiments and the recovery time for environmental conditions; subsequently a new group of turtles was placed into the observation area in order to repeat the simulation.

Experiments
Determination of temperament of Rhinoclemmys areolate. It was necessary to identify stable behavioral traits over time and in different situations. For this aim,16 males were observed individually and in isolation in four different situations: • Reaction to manipulation: a male is removed from confinement and handled briefly for 1 minute.
• Reaction to risk of predation: a male is removed from confinement, handled briefly and gently released in an overturned position, that is, on his carapace.   • Reaction to a new environment: a male is taken from the confinement and released into the experimental area unknown to the individual. (30min) • Reaction to the introduction of a new object: a male is removed from confinement, released into the experimental area and, after 5 minutes, an unknown object is suddenly presented (stuffed dog-like animal). (30min) In order to organize males according to their level of boldness-shyness we: (i) extrapolated behavioral events of each simulation from video recordings, (ii) converted them into variables, (iii) defined occurrence (= 1) and absence (= 0) of each event obtaining binomial values; (iv) defined frequency and (v) time in seconds when possible, obtaining quantitative values; (vi) for the latter, the moment an abrupt change of slope occurred within the data range was recorded, allowing the classification of males as shy or bold (S1 Table).
Each event was considered as a variable and inserted into an Excel sheet. A boldness index was given to each male, dividing the sum of "occurrences" of variables that could describe a bold temperament by the total number of variables. This analysis allowed us to classify individuals into groups, according to their degree of boldness.
Subsequently, variables were analysed using a correlation matrix and the obtained values were inserted into a multidimensional scaling (MDS) that enabled the creation of a map showing relative positions between different events and relative distances.
Finally, a Mann-Whitney test was applied to determine which variables were significant in determining temperament.
Before initiating interaction experiments, we tested the influence of turtle length (that is considered a good indication of the age of each individual) on temperament. We implemented this test to ensure that our results were not the consequence of turtle size or age. Both groups obtained were analysed in relation to straight carapace length (SCL) and straight plastron length (SPL) by applying a Mann-Whitney test (S2 Table).
Temperament as a modulating factor during the interaction between male and female. 16 groups of turtles consisting of one male and one female were created randomly. Each couple was released, sequentially, within the experimental area, each individual inside a metal box. They were soaked with water using an artificial sprayer to promote reproductive behavior and later freed to move around for 30 minutes.
For this experiment, the same individuals were not used more than once. Each group remained sequentially in the experimental arena and was carefully observed. As for the first simulations, we (i) extrapolated behavioral events from each simulation, (ii) converted them into variables, (iii) defined occurrence (= 1) and absence (= 0) of each event obtaining binomial values; (iv) defined frequency and (v) time in seconds when possible, obtaining quantitative values; (vi) for the latter, the moment an abrupt change of slope occurred within the data range was recorded (S3 and S4 Tables).
Each event was considered as a variable and inserted into an Excel sheet. Variables were analysed using a correlation matrix and the obtained values were inserted into a multidimensional scaling (MDS) allowing the creation of a map showing relative positions between different events and relative distances.
Temperament as a modulating factor during the competitive interaction between males in the presence of a female. Eight experimental groups were randomly created, each consisting of two males and one female. For this experiment, the same individuals were only used once.
The individuals were placed equidistant within the experimental area, soaked with water using an artificial sprayer to promote reproductive behavior and released to move around for 30 minutes.
Each group remained in the experimental area and was carefully observed. As for the first simulations, we (i) extrapolated behavioral events from each simulation, (ii) converted them into variables, (iii) defined occurrence (= 1) and absence (= 0) of each event obtaining binomial values; (iv) defined frequency and (v) time in seconds when possible, obtaining quantitative values; (vi) for the latter, the moment an abrupt change of slope occurred within the data range was recorded (S5 and S6 Tables).
Each event was considered as a variable and inserted into an Excel sheet. Variables were analysed using a correlation matrix and the obtained values were inserted into a multidimensional scaling (MDS) that allowed the creation of a map showing relative positions between different events and relative distances.

Determination of temperament of Rhinoclemmys areolata
Of 17 behavioral traits observed (S1 Table), we selected 12 for study (Table 1). For each male, the presence or absence of each of these traits was defined (Table 2). For this study, we used "bolder" and "shier" terms instead of "bold" and "shy" since boldness is a continuous spectrum and it is not possible to create sharp and fixed fractures for group definition.
In order to create two groups for the following experiments, we considered all turtles with an index less than or equal to 0.5 as shier, while those with an index greater than 0.5 bolder. As a result, individuals were divided into 9 bolder and 7 shier males ( Table 2). Straight carapace length and straight plastron length did not show a significant relationship with temperament (Carapace: Mann-Whitney U test: U = 20, P = 0.25; Plastron: Mann-Whitney U test: U = 19, P = 0.21) (S7 Table).
Rearranging the variables according to presence, frequency and times (Table 1), and through the multidimensional scaling, we obtained a total of 10 variables: five correlated with the shiest individuals and five associated with the boldest individuals (Fig 5).
These variables were characteristic of the shier turtles: (i) hiding in a new environment, (ii) not crossing a new environment, (iii) a high duration of quiet time in a new environment, (iv) high straightening time during predation, and (v) low walking time in a new environment. Variables for boldest turtles were: (i) not hiding in a new environment, (ii) cross a new environment, (iii) low quiet time in a new environment, (iv) low straightening time during predation, and (v) high walking time in a new environment (Fig 5). Four of those variables were also statistically significant for the discrimination of the two groups (Fig 6, S8 Table).

Temperament as a modulating factor during the interaction between male and female
As results for the relation between temperaments and male and female interaction, we observed behaviors linked with courtship and others with mating.
Regarding courtship, we determined 16 behaviors (S3 Table), but we only used 11 for the analysis, excluding behaviors showed by most of the individuals and those that did not present any differences between groups.
We determined if the behaviors occurred or not and how often to obtain their frequency (low or high) ( Table 3). We observed that the following traits corresponded to shier individuals (Fig 7): (i) pushing the female, (ii) blocking the female, (iii) frontal approach with stretched legs and retracted neck (iv) pushing the ground with front leg, (v) approaching when excited. On the other hand, some bolder turtles did not show courtship behavior while in others this was not very frequent. Bolder individuals displayed the following behaviours: (i) do not block Table 2

PLOS ONE
Temperament and sex in a terrestrial turtle the female, (ii) do not push it, (iii) do not approach excitedly, (iv) do not walk backwards, (v) low frequency of frontal approach with stretched legs and extended neck (Fig 7). Four behaviors (Table 4, S4 Table) were observed for mating. None of these behaviors was displayed by the shiest individuals, while the boldest were more associated with intent of mounts, and mount at high frequency (Fig 8, S1 Video).

Temperament as a modulating factor during the competitive interaction between males in the presence of a female
We obtained 12 behaviors related to courtship (Table 5, S5 Table). We considered the temperament of the two males together with a female and observed that courting variables are concentrated around the boldest regardless the temperament of their opponent (Fig 9). In contrast, the shiest individuals did not display particular behaviors with the exception of blocking the female during tracking (Fig 9) (S2 Video).

Discussion
A variety of methods were used separately to determine temperament in aquatic and terrestrial turtles [4,[41][42][43][44][45]. This is the first paper that tests all of these methods. In this study, we tested reaction to: manipulation, predation risk, a new environment and the introduction of a novel object with the aim of finding consistency, to determine accurately behaviors and in consequence the best method to define the level of boldness in turtles. Behaviors observed in a new environment appear to be the most reliable for determining boldness in turtles: hiding and limited walk for the shiest, quiet (shiest) or not quiet (boldest), walking frequently and crossing the new environment for the boldest. These results coincide with two previous works [43,44].
The test with the risk of predation also demonstrates that the time taken to straighten is associated with boldness, as observed in previous studies [41,42].
Our work focused only on males, and determining temperament in females or juveniles may be different even though the exploration of a new environment may be comparable irrespective of the sex and age of turtles that generally share the same life strategy. Therefore, to assure that behaviors associated with the reaction to a new environment can provide an improved method, we suggest further studies that examine this in females and juveniles.
An interesting element that emerges in this study is the modulation of males´behavior. When a bold male is alone with a female, he often skips courtship behaviors to devote himself to stimulating the female for mounting and mating; when the same male is found in the presence of another conspecific male, a potential sexual competitor, he first engages in courtship behaviors towards the female. A similar result has been observed in the common water strider Aquarius remiges (Gerridae, Insecta), where the most active and aggressive males spend most of their time searching for a female as well as presenting the highest copulation rates [52]. In the common lizard Zootoca vivipara (Lacertidae, Reptilia), males' behavior together with the predatory context influence the mating preference of the females. Those that have not been exposed to predatory signals prior to mating, mate more often with more active males. If, on the other hand, these have been subjected to predatory signals, they tend to choose less active males that are more likely to have a longer life expectancy and to avoid the more active males in order to increase the survival of their offspring in an environment with risk of predation [53]. In our simulations, mating attempts of shier males were very rare; when the shier turtle is in presence of a competitor and a female, it ignores the female. In summary, the boldest individuals court females, but only if another male is present; otherwise these males immediately mount and attempt to copulate. Shier males will only court if other males are absent; even then mating is rare. This kind of behavior in the presence of a competitor was observed in Mauremys leprosa where responses to chemicals cues depended on boldness; in particular, shy turtles avoided chemical cues from familiar and unfamiliar bolder males [41]. However, it is possible shier animals require more time to display courtship and mating behavior. Therefore, we can argue that the degree of boldness can influence males' reproductive behavior, in decision making and their reaction according to the situation. In terms of costs and benefits, boldness can be considered a factor that allows individuals to evaluate how to act during courtship and mating considering the presence of conspecifics of both sexes. Shier turtles could avoid injuries and the possibility of losing energy when a bolder male is competing. In our

PLOS ONE
approach the same female. In other species, male competition could be very costly because of fights, injuries, territory care and predatory danger during courtship. Some reptiles avoid or reduce their response to such competition and could provide an alternative strategy to prevent damage and predator costs, as shown in the Iberian wall lizard Podarcis hispanica, which uses chemical cues during competitive matches to recognize competitors and reduce fight intensity and resulting costs [54]. The success of different strategies dependent on temperament is to some extent subject to environmental constraints such as the presence of predators as observed for other animals [1]. For example, the broadhead skink Eumeces (Plestiodon) laticeps shows a sensitivity in predation risk during reproductive opportunities: when females were present, isolated and mate-guarding males initiate courtship, but isolated males permitted closer approach by a predator than did mate-guarding males. Furthermore, there is a high level of latency displayed by isolated males when a predator is close to the female, allowing them to balance predation risk with courtship behavior [55].
Researchers have already discussed that behavior plays a key role in animal survival and evolution, behavioral traits can help species to adapt to rapid changes in the environment [56], and survive to natural selection [57] as much as morphological and physiological traits. Differences in sexual behavior related to boldness or other behavioral traits can be an honest signal of mate quality [58]. Therefore, identifying these behavioral traits that can affect population viability is important to provide helpful information to integrate conservation plans [59].
The information obtained in this study represents an important tool for the conservation of reptiles and in particular for land turtles. Worldwide, turtles represent one of the vertebrate groups with the highest risk of extinction [60]. In recent years, many researchers have ensured the importance of not only increasing the behavioral information of the species, but also implementing these data in their conservation and monitoring strategies [14,61].
Assumptions have been made in the past about the likely relationship between temperament and reproductive behaviors [62], but to our best knowledge, this is the first work that formalizes it experimentally for land turtles.
The achievements in this study can be used as a first step for compatibility plans for turtles, in order to combine temperament traits that may contribute to successful breeding couples [14].
Temperament studies in couple compatibility have significant implications in ex situ conservation programs, since specific traits (such as aggressiveness and fear) have been related to reproductive success in some species raised in zoos [15]. In addition, certain temperaments may be more likely to thrive in captive environments, artificial selection for or against specific temperament traits may accelerate domestication processes, making animals raised for conservation less suitable for release in wildlife [32].
Our research showed that bolder males need less time to mount and mate, with perhaps higher fitness than shier ones.
Another application of our findings is the possibility of conserving the whole range of boldness. Considering behavioral variation, it is important to preserve boldness since it allows conservationists to reintroduce animals with all behavioural traits, giving them proper tools to survive in novel environments [57]. Considering this, in order to not to lose the behavioral pool during captive reproduction, we must be aware of the needs of different individuals. Therefore, our study demonstrates that shy turtles probably need more time for mating and need to be alone with the female, otherwise they lose interest. For the boldest animals, the necessary times are lower and they are potentially able to reproduce even with the presence of conspecifics.
Supporting information S1 Table. List of responses found in behavioral simulation (occurrence = 1, absence = 0), with relative abbreviation and subdivision applied for statistical analysis. The first column shows the names of each individual, bolder individuals have a grey backgrounds while those considered shier have white.